Check results from same check command but diffrent service.

mguthrie · Post by **mguthrie** » Wed Jul 14, 2010 11:18 am

Any chance we could see the actual performance data in a text file so we can check it's syntax?

Post by **Box293** » Thu Jul 15, 2010 5:44 am

Sure thing. I've attached a zip that contains .rrd and .xml files for both a working and non working service.

Screenshots as well.

mmestnik · Post by **mmestnik** » Thu Jul 15, 2010 2:01 pm

We are looking for the performance data from the advanced page of the service. I'd like a text copy/past so that I can review and test the data. Please include the check output as well. If you like you can save the HTML source.

Post by **Box293** » Fri Jul 16, 2010 9:35 am

No problems.

It now seems that all hosts are displaying nan for this service.

If I look at the graph for the past week you can see two day's worth of data and then nothing.

mmestnik · Post by **mmestnik** » Mon Jul 19, 2010 10:29 am

Ok, now we got the problem. There is an unclosed string the error might look like this "Expected \' but got EOF."

Code: Select all

'Terminal Services Active Sessions is %.f'=0.000000%;0.000000;0.000000;

The other issue I see is that "%.f" could be a formatted print function that's not being resolved. The third and final issue is that RRD and thus pnp can only graph integers, so what should be being printed here is %i, not %f or %.f anything. These numbers should not have '.' they need to be whole.

This information is provided by the check command you are using, this information can be found in the CCM. That screen would also be helpful, it's nsclinet++ no? So then the nsclient++ application has some configuration related to the config in CCM that's important. That should finally bring us to the application that needs to be reprogrammed.

I'll ask about what we should be doing with bugs in nsclient++ plugins, bugs in nsclient++ we would defiantly handle as we would like to maintain a good professional relationship with that team. The plugins on the other hand it might be better for you to seek assistance directly from the author or team responsible, unless it's the nsclient++ team of course.

Post by **Box293** » Tue Jul 20, 2010 8:11 am

mmestnik wrote: This information is provided by the check command you are using, this information can be found in the CCM. That screen would also be helpful, it's nsclinet++ no? So then the nsclient++ application has some configuration related to the config in CCM that's important.

Yes, I am using NSClient++ 0.3.7.493

Here's a screenshot of the service settings.

Service Settings.png

The code for $ARG3$ is:

Code: Select all

-l "\\Terminal Services\\Active Sessions","Terminal Services Active Sessions is %.f"

So the whole command is (excluding -s):

Code: Select all

check_nt -H <hostname> -p 12489 -v COUNTER -l "\\Terminal Services\\Active Sessions","Terminal Services Active Sessions is %.f"

When run from the cli I get:

Code: Select all

Terminal Services Active Sessions is 2 | 'Terminal Services Active Sessions is %.f'=2.000000%;0.000000;0.000000;

mmestnik wrote:
Code: Select all
'Terminal Services Active Sessions is %.f'=0.000000%;0.000000;0.000000;
The other issue I see is that "%.f" could be a formatted print function that's not being resolved. The third and final issue is that RRD and thus pnp can only graph integers, so what should be being printed here is %i, not %f or %.f anything. These numbers should not have '.' they need to be whole.

I try with %i:

Code: Select all

check_nt -H <hostname> -p 12489 -v COUNTER -l "\\Terminal Services\\Active Sessions","Terminal Services Active Sessions is %.i"

When run from the cli I get:

Code: Select all

Terminal Services Active Sessions is 0 | 'Terminal Services Active Sessions is %i'=2.000000%;0.000000;0.000000;

This seems to report the number of active sessions as 0 instead of 2, however it does still report 2 in the final section.

Also, I try with %f (instead of %.f):

Code: Select all

check_nt -H <hostname> -p 12489 -v COUNTER -l "\\Terminal Services\\Active Sessions","Terminal Services Active Sessions is %f"

When run from the cli I get:

Code: Select all

Terminal Services Active Sessions is 2.000000 | 'Terminal Services Active Sessions is %f'=2.000000%;0.000000;0.000000;

And this reports the number with a bunch of 0's after the decimal point.

Just as a comparision, the supplied Windows Server Configuration Wizard will create the following three Performance Counter service definitions:

Code: Select all

check_nt -H <hostname> -p 12489 -v COUNTER -l "\\Server\\Errors Logon","Login Errors since last reboot is %.f" -w 2 -c 20
Login Errors since last reboot is 0 | 'Login Errors since last reboot is %.f'=0.000000%;2.000000;20.000000;

check_nt -H <hostname> -p 12489 -v COUNTER -l "\\Paging File(_Total)\\% Usage","Paging File usage is %.2f %%" -w 70 -c 90
Paging File usage is 0.17 % | 'Paging File usage is %.2f %%'=0.166791%;70.000000;90.000000;

check_nt -H <hostname> -p 12489 -v COUNTER -l "\\Server Work Queues(0)\\Queue Length","Current work queue (an indication of processing load) is %.f " -w 4 -c 7
Current work queue (an indication of processing load) is 0  | 'Current work queue (an indication of processing load) is %.f '=0.000000%;4.000000;7.000000;

You can see my example is very similar to these ones, mine don't specify warning or critical values. This is why I am confused about getting the nan issue.

mmestnik wrote:These numbers should not have '.' they need to be whole.

The values above indicate the following:

%f gives a number with 6 digits after the decimal point
%.f gives a whole number
%.1f gives a number with 1 digit after the decimal point
%.2f gives a number with 2 digits after the decimal point
and so on ...

Let me know any other information you need.

On a side note, it seems that the graphing system uses a % sign as the default Y axis label if it cannot determine what it is. You can see this when you look at a Logon Errors since last reboot graph. It would be good to be able to specify the label.

mmestnik · Post by **mmestnik** » Tue Jul 20, 2010 11:43 am

Great job on digging down to all this information.
Now I'm just all confused by what this would indicate.

After talking with some folks it seams the legacy limitation of RRDtool to only handle integers is gone, it seams to use floats now. Also it looks like PnP will mangle the data, converting seconds into milliseconds(if it understands the unit). To this I replayed, so if s stood for "snails" and not seconds there would be a problems. At the time I just couldn't think of something like [1]saros.

I don't have anything currently, let me ponder on this for a bit.

1. http://www.unc.edu/~rowlett/units/dictS.html

Nagios Support Forum

Check results from same check command but diffrent service.

Re: Check results from same check command but diffrent service.

Re: Check results from same check command but diffrent service.

Re: Check results from same check command but diffrent service.

Re: Check results from same check command but diffrent service.

Re: Check results from same check command but diffrent service.

Re: Check results from same check command but diffrent service.

Re: Check results from same check command but diffrent service.